(NLP Pipeline with NLTK + Scikit-learn)

This project applies Natural Language Processing and Machine Learning to classify text statements into emotional categories like Anxiety, Depression, Relief, and more. It uses a combination of NLTK for preprocessing and scikit-learn for modeling and evaluation.

This project uses the dataset:
🔗 Sentiment Analysis for Mental Health – Kaggle

• Contains thousands of labeled statements collected from mental health-related sources.
• Each entry has a statement (text) and a status (emotion category).
• The dataset focuses on mental health sentiment and reflects a wide range of human emotions like:
  • Anxiety
  • Depression
  • Loneliness
  • Optimism
  • Gratitude
  • Relief

This makes the dataset ideal for building models that help understand emotional expression in real-world mental health contexts.

1. Prepares and cleans text using NLTK tokenization and stopwords.
2. Converts text into numerical features using CountVectorizer.
3. Trains a Linear Support Vector Classifier (LinearSVC).
4. Evaluates the model using accuracy, precision, recall, F1-score.
5. Uses RandomizedSearchCV to optimize the C hyperparameter.
6. Outputs the best model and its performance metrics.

• There are learnable patterns in the language used to describe emotional states.
• Even a simple model like LinearSVC achieves ~75% accuracy, showing that:
  • Words and phrases strongly correlate with specific emotions.
  • Emotional text can be quantified and predicted with solid performance.
• The model generalizes well across multiple emotional labels, especially after tuning C to 0.1.

In other words:

The language people use when expressing mental health concerns contains enough signal for a machine learning model to recognize and classify emotions with meaningful accuracy.

nltk.download('punkt')
nltk.download('stopwords')

data = pd.read_csv('Combined Data.csv')
X = data['statement'].fillna("").astype(str)
y = data['status']

vectorizer = CountVectorizer()
X_train_features = vectorizer.fit_transform(X_train)

clf = LinearSVC()
clf.fit(X_train_features, y_train)

accuracy_score(y_test, y_pred)
precision_score(...)
confusion_matrix(...)

RandomizedSearchCV(..., param_distributions={'C': [...]})

To run the code:

1. Download the dataset from Kaggle.
2. Install required libraries:

pip install nltk scikit-learn pandas

3. Upload the CSV file and run the notebook.